Room Temperature Electroluminescence in Si:
Rare-Earth Impurities
There has been an enormous amount of work trying to obtain an intense luminescence signal from Si and the recent observations of room temperature photoluminescence (PL) [1, 2] and, even more remarkably, room temperature electroluminescence in Si:Er [3] have created great interest in the materials science community.
The intra-4f electronic transitions in rare-earth (RE) impurities
involve very localised atomic orbitals and their absorption and
luminescent bands are extremely sharp and characterised by long
lifetimes even in disordered environments such as glasses
[4, 5]. Er has been a favoured dopant in Si since its
emission wavelength is a minimum for both absorption and dispersion in
fibre-optic cables. The work on Si is particularly important as the
mature processing technology can be utilised in the development of
optical communication devices. The 
I 
I transition ( 
1.54 
m) is dipole forbidden
in the atom but is allowed through wave-function overlap with the
ligands. The MIT and AT&T groups have shown that co-doping with O, or
other light elements, increases the PL intensity by factors up to 80:
there is only weak PL in float zone Si implanted with Er
[6]. It is the understanding of this remarkable
increase of the PL intensity with co-doping that is the main aim of
this project. Although experimental work is well advanced, there has
been very little theoretical work into this effect or even into the
theory of RE impurities in semiconductors.
There is then a timely need to explore theoretically the effects due to the environment, especially in regard to gap levels, the covalent and crystal field contributions to the f-level splitting as well as the intra-f dipole matrix elements. These are the principal aims of the project.
A current technical problem is the low power output arising from the
long radiative life-time and the low concentration of optically active
Er defects. One way of getting around the former may be by implanting
REs in nano-crystalline Si which in principle could lead to a
large increase in power. For example, Mn in nano-crystalline ZnS has a
radiative life-time 10 
times that of Mn in bulk ZnS
[7]. This effect is probably due to a confinement of the
exciton to the nano-crystal. These ideas will be explored in the
proposal.
The environment of the rare-earth (RE) impurity in the lattice for four effects. Firstly, in the introduction of gap levels without which the the first stage of the PL, namely the trapping of an e-h pair would be inefficient. However, evidence for RE induced gap levels is scanty. DLTS studies of Si:Er suggest some [8] although no such levels are seen when MBE grown material was used [9] duue to segregation of the RE [10, 11]. Gap levels are found in DLTS studies of InP:Yb [12]. Secondly, a crystal field or more likely, covalent bonding [13] with the RE's neighbours, split the f-levels leading to terms of odd parity in the potential and hence to non-zero dipole elements permitting the optical transition. Thirdly, a strong RE-neighbour interaction leads to low diffusivity of the RE and the inhibition in the growth of a metallic phase. Finally, O can cause RE ions to switch sites from interstitial to substitutional possibly leading to optical activity.
Information, however, on the lattice location and neighbours of the RE
is notoriously difficult to find. EXAFS has given some details for
glasses [14, 15, 16] and there has very recently been
some RBS data [17] in the compound semiconductors. The
conclusion is that, except for Yb in InP, RE ions in III-V materials
and Er in Si [18] do not occupy lattice sites. Two
problems are, however, that the optically active fraction of the RE
defect might occupy a site distinct from the major inactive species
and that co-doping causes site switching. Takahei et al [19]
concluded from channelling and SIMS studies on O-doped GaAs:Er that Er
was located at a Ga site with two O atoms at neighbouring As sites.
Thus the RBS data indicates that O drives Er on-site. Other PL
studies [20] conclude that Er is at a 
interstitial
site in FZ-Si but in Cz-Si, Er defects with lower symmetry are found.
Given the extreme difficulty of determining the active sites of the rare earth impurity, there has been interest in applying ab initio modelling techniques.